1
Neuroscience
J Physiol 000.00 (2015) pp 1–14
Sleep spindles and human cortical nociception: a surface and intracerebral electrophysiological study L´ea Claude1 , Florian Chouchou1 , Germ´an Prados1 , Ma¨ıt´e Castro1 , Barbara De Blay1 , Caroline Perchet1 , Luis Garc´ıa-Larrea1 , St´ephanie Mazza2 and H´el`ene Bastuji1,3 1
Central Integration of Pain (NeuroPain) Lab – Neuroscience Research Center, INSERM U1028, CNRS UMR5292, Lyon, France Universit´e Lumi`ere Lyon 2, Laboratoire d’Etude des M´ecanismes Cognitifs (EMC), Bron, France 3 ´ Unit´e d’Hypnologie, Service de Neurologie Fonctionnelle et d’Epileptologie, Hˆopital Neurologique, Hospices Civils de Lyon, Bron, France 2
The Journal of Physiology
Key points
r Sleep spindle are usually considered to play a major role in inhibiting sensory inputs. r Using nociceptive stimuli in humans, we tested the effect of spindles on behavioural, autor r r
nomic and cortical responses in two experiments using surface and intracerebral electroencephalographic recordings. We found that sleep spindles do not prevent arousal reactions to nociceptive stimuli and that autonomic reactivity to nociceptive inputs is not modulated by spindle activity. Moreover, neither the surface sensory, nor the insular evoked responses were modulated by the spindle, as detected at the surface or within the thalamus. The present study comprises the first investigation of the effect of spindles on nociceptive information processing and the results obtained challenge the classical inhibitory effect of spindles.
Abstract Responsiveness to environmental stimuli declines during sleep, and sleep spindles are often considered to play a major role in inhibiting sensory inputs. In the present study, we tested the effect of spindles on behavioural, autonomic and cortical responses to pain, in two experiments assessing surface and intracerebral responses to thermo-nociceptive laser stimuli during the all-night N2 sleep stage. The percentage of arousals remained unchanged as a result of the presence of spindles. Neither cortical nociceptive responses, nor autonomic cardiovascular reactivity were depressed when elicited within a spindle. These results could be replicated in human intracerebral recordings, where sleep spindle activity in the posterior thalamus failed to depress the thalamocortical nociceptive transmission, as measured by sensory responses within the posterior insula. Hence, the assumed inhibitory effect of spindles on sensory inputs may not apply to the nociceptive system, possibly as a result of the specificity of spinothalamic pathways and the crucial role of nociceptive information for homeostasis. Intriguingly, a late scalp response commonly considered to reflect high-order stimulus processing (the ‘P3’ potential) was significantly enhanced during spindling, suggesting a possible spindle-driven facilitation, rather than attenuation, of cortical nociception. (Received 14 May 2015; accepted after revision 23 August 2015; first published online 2 September 2015) Corresponding author H. Bastuji: Unit´e d’Hypnologie, Service de Neurologie Fonctionnelle et d’Epileptologie, Hˆopital Neurologique, Hospices Civils de Lyon, 59 Boulevard Pinel, 69500 Bron, France. Email:
[email protected] Abbreviations AASM, American Academy of Sleep Medicine; EKG, electrocardiogram; EMG, electromyogram; EOG, electrooculogram; LEP, laser-evoked potentials; MNI, Montreal Neurological Institute; MRI, magnetic resonance imaging; REM, rapid eye movement; SEEG, stereo-electroencephalography.
C 2015 The Authors. The Journal of Physiology C 2015 The Physiological Society
DOI: 10.1113/JP270941
2
L. Claude and others
Introduction First described by Loomis et al. (1935), spindles are transient cerebral activities occurring during non-rapid eye movement (REM) sleep, and appearing on EEG as ‘waxing and waning’ 12–15 Hz oscillating waves lasting 0.5-2 s (De Gennaro & Ferrara, 2003). According to model of Steriade (2006), spindling is initiated at the cellular level by rhythmic inhibitory oscillations at 12–15 Hz in thalamic reticular gabaergic neurons, which induce recurrent postinhibitory rebound spike bursts in thalamocortical glutamatergic units. These bursts lead cortical neurons to respond at spindle’s frequency. Consecutively, cortical feedback to the thalamus synchronizes this oscillation inside the entire thalamocortical network (De Gennaro & Ferrara, 2003; Astori et al. 2013). Despite considerable knowledge about the mechanisms underlying the generation of spindles, their function remains poorly understood and to some extent controversial (Astori et al. 2013). Pioneering work by Yamadori (1971), suggested a role as ‘sensory gate’, useful to preserve sleep by inhibiting sensory input. It has been reported that individuals who generate more spindles also have greater tolerance to noise during a noisy night of sleep (Dang-Vu et al. 2010) and that both evoked potentials and haemodynamic (blood oxygen level-dependent) thalamocortical responses become attenuated when pure-tone auditory stimuli are delivered during spindle activity (Elton et al. 1997; Cote et al. 2000; Dang-Vu et al. 2011; Schabus et al. 2012). However, such an inhibitory role of sleep spindles has not been supported universally, and Moruzzi et al. (1950) were the first to state that ‘several types of evoked electro-cortical activities [ . . . ] undergo pronounced augmentation during spindle bursts’. In this respect, two studies in humans failed to show any blockade of sensory or cardiovascular responses to auditory stimuli during spindling (Church et al. 1978; Crowley et al. 2004), and one of them even suggested that sleep spindles may reflect ‘phasic reductions in inhibitory action’, resulting in increased transmission of sensory events (Church et al. 1978). In the intact cortex of cats, spontaneous spindles were shown to induce increased synaptic responsiveness to single stimuli, suggesting that they might actively induce neural plasticity (Timofeev et al. 2002). Thus, the exact role of sleep spindling in the sensory inputs processing remains controversial. Sleep spindling might have different cortical actions depending on the type, or intensity, of the stimulus received. For example, stimuli within the frequency range of sleep spindles triggered ‘augmenting’ cortical responses in slabs of cat’s cortex when their intensity was relatively high, whereas, at low intensities, cortical responses were decremental (Timofeev et al. 2002). Hence, some of the discrepancies reported previously might arise from the use of stimuli with dissimilar behavioural
J Physiol 000.00
relevance. Nociceptive stimuli are included among those with highest relevance for survival and have a six-fold greater probability of awakening the sleeper than the auditory tones used in most previous studies (Lavigne et al. 2004; Bastuji et al. 2008); thus, they might provide a straightforward demonstration of the effect of spindles on sleep disruption by external inputs. Also, in contrast to other equally relevant stimuli, nociceptive pulses can be made sufficiently short, in the order of milliseconds, and easily included within or outside the duration of a spindle. The present study aimed to obtain a comprehensive description of the modulation produced by spindle activity. Accordingly, we assessed behavioural, cortical and autonomic reactions during spindling using phasic nociceptive thermal stimuli. Two all-night experiments were conducted: one with surface EEG recordings in healthy subjects and the other using intracerebral recordings in epileptic patients, in whom detection of spindles within the thalamus was coupled with the recoding of sensory responses in the posterior insula, which is the sensory region responding most systematically to nociceptive stimuli (Garcia-Larrea & Peyron, 2013). Methods Subjects
The total sample explored comprised 18 subjects: nine healthy volunteers and nine epileptic patients with implanted intracerebral electrodes. The two experiments were approved by the local Ethics Committee (CCPPRB L´eon B´erard-Lyon) and were supported by the French National Agency for Medical Research (INSERM). Healthy subjects. The nine healthy volunteers (six men,
mean ± SD age 30.2 ± 7.4 years) were free of neurological, psychiatric, chronic pain or sleep disorders and were not receiving any psychotropic medication. All subjects provided their informed consent to take part in the experiments, which were conducted in accordance with the Declaration of Helsinki. Subjects were also paid for their participation.
Patients with intracerebral implanted electrodes. The nine patients included in the study (six men, mean ± SD of age 30.9 ± 11.3 years) suffered from partial pharmacoresistant epilepsy. To delineate the extent of the cortical epileptogenic area and to plan a tailored surgical treatment, depth EEG recording electrodes (diameter 0.8 mm; 5–15 recording contacts 2 mm in length, intercontact interval 1.5 mm) were implanted perpendicular to the mid-sagittal plane, in accordance with the stereotactic technique of Talairach & Bancaud (1973). The decision to explore specific areas resulted from the observation C 2015 The Authors. The Journal of Physiology C 2015 The Physiological Society
J Physiol 000.00
Sleep spindles do not inhibit pain responses in humans
during scalp video-EEG recordings of ictal manifestations suggesting the possibility of seizures either propagating to or originating from these regions (Guenot et al. 2001). The thalamus, at or near the pulvinar region, was one of the targets of stereotactic implantation, which, as a result of its reciprocal connections with temporal and parietal cortical areas, may be involved in most of temporal and insular lobe seizures (Rosenberg et al. 2009; Bastuji et al. 2015). Simultaneous exploration of the thalamus and neocortical areas was possible using a single multicontact electrode, such that thalamic exploration did not increase the risk of the procedure by adding one further electrode track specifically devoted to it. In agreement with French regulations relative to invasive investigations with a direct individual benefit, patients were fully informed with respect to the electrode implantation, stereotactic EEG, evoked potential recordings and cortical stimulation procedures used to localize the epileptogenic cortical areas and all provided their informed consent. Stimuli
Radiant nociceptive heat pulses of 5 ms in duration were delivered with a Nd:YAP laser (yttrium aluminium perovskite; wavelength 1.34 lm; ElEn, Florence, Italy). The laser beam was transmitted from the generator (outside the bedroom) to the stimulating probe via an optical fibre of 10 m in length (550 µm in diameter with subminiature version A-905 connector; Amphenol Fiber Optic Products, Lisle, IL, USA). Series of laser stimuli were delivered on the dorsum of the right or left hand, alternatively, in healthy subjects, and contralateral to the hemisphere of electrode implantation in patients. The intensity of the laser pulses was kept stable for any given subject during the whole night, slightly above the individual pain threshold obtained at wake. This threshold corresponded to a level of 4–5 on a verbal numerical scale ranging from 0 to 10 (where 0 = no sensation and 10 = unbearable pain; with the intermediate levels being: 1 = barely perceived; 2 = lightly pricking, not painful; 3 = clearly pricking, not painful; 4 = barely painful, like pulling a hair; 5 = painful, prompting to rub the skin; 6 = very painful and distressing; 7 and more = strongly unpleasant pain). Pain thresholds were obtained in all subjects with energy densities of 50–79 mJ mm–2 . These pain threshold values were within the normal range of data classically obtained in our laboratory, and are in accordance with the reported experimental data using Nd:YAP lasers (Leandri et al. 2006; Perchet et al. 2008). To avoid damaging the skin, habituation and peripheral nociceptor fatigue, stimulus blocks consisted of a maximum of 20 laser pulses and the heat spot was slightly shifted over the skin surface after each stimulus (Schwarz et al. 2000). The targeting of laser and the slight repositioning were carried out manually, keeping a stable distance to obtain a 4 mm spot, and C 2015 The Authors. The Journal of Physiology C 2015 The Physiological Society
3
slightly moving the spot between two stimuli within the hand dorsum, in the territory of the superficial branch of the radial nerve (Cruccu et al. 2008). Because preliminary studies showed that delivering stimuli at short ( 15 s) after a laser stimulus (F1,17 = 4.74; P = 0.0439; cortical arousal: 22 ± 4.1% and awakening: 10 ± 4.0%). This phenomenon appeared to be independent of the presence of a spindle simultaneously to the nociceptive stimulation because the interaction between the two factors was not significant (F1,17 = 1.71; P = 0.21).
J Physiol 000.00
Laser-evoked responses Scalp recordings. In the surface experiment, the sensory N2–P2 responses and the late positivity following P2, labelled P3 wave, were considered. No significant difference was detected between the S and NS stimulus conditions for N2 and P2 latencies (N2: F1,8 = 0.07; P = 0.8; P2: F1,8 = 1.69; P = 0.23), nor for N2–P2 amplitude (F1,8 = 1.74; P = 0.22). A significant effect of electrode position (topography) was observed on the N2–P2 amplitude (F2,16 = 11.26; P = 0.0009), which was smaller on frontal electrodes than on central (t16 = 4.07; P = 0.0027) and parietal recordings (t16 = 4.15; P = 0.0023) (Fig. 4 and Table 1). These topographical changes were independent of the existence or not of a concomitant spindle, as shown by the absence of interaction between topography and spindle condition (F2,16 = 0.71; P = 0.51) (Fig. 4 and Table 1). The latency of the P3 component was not significantly different in the S and NS stimulus conditions (F1,8 = 1.13; P = 0.32). Conversely, the P3 amplitude was significantly different between the two conditions, and higher when the stimulus was delivered during a spindle (S condition:
Autonomic responses
Two-way ANOVA showed that mean cardiac RR interval was significantly modified as a function of time (F1,8 = 23.05; P = 0.0014) but not of the stimulus condition (S/NS) (F1,8 = 0.03; P = 0.61), with no interaction (F1,8 = 0.55; P = 0.48). The mean cardiac RR interval was significantly decreased) (i.e. heart rate increase, following the nociceptive stimuli compared to pre-stimulus period; before: 1139 ± 31 ms; after: 992 ± 34 ms) (Fig. 3).
70
Stim
Stim
1200
65
1100
Heart rate (bpm)
RR intervals (ms)
1150
1050
60
55
1000
Condition: ns Time: p10 s vs. 3 s). In our case, this longer interstimulus interval allowed an untying the effect of spindle from that of habituation related to the repetition of the stimuli. Also, suprathreshold nociceptive stimuli at rates faster than 0.2 Hz tend to create wind-up phenomena, already arising at spinal level, which would have complicated the interpretation of results. Such a discrepancy between auditory studies and the C 2015 The Authors. The Journal of Physiology C 2015 The Physiological Society
11
results of the present study may also relate to the unique homeostatic relevance of nociception, where alerting capacities are essential for survival. Differences in sensory systems carrying noxious and non-noxious information are particularly apparent in their differing GABAergic circuitry within the thalamus. The synaptic relationships of non-nociceptive thalamic terminals take the form of ‘triads’, whereby the ascending axon simultaneously activates thalamocortical relay neurons and the dendritic appendages of inhibitory GABA interneurons (Ralston & Ralston, 1994). However, this arrangement, considered to mediate feed forward inhibition of thalamocortical cells (Ohara & Lieberman, 1993), is absent in more than 85% of spinothalamic afferents, which form simple axodendritic synapses with relay cells, and do not contact inhibitory GABA interneurons (Ralston & Ralston, 1994). Such a simple circuitry suggests that the transmission of noxious information is much less subject to GABAergic interneuronal modulation than non-noxious information carried by the lemniscal afferents (Ralston & Ralston, 1994; Ralston, 2005). In anaesthetized rats, nociceptive (but not tactile) input was able to inhibit a significant portion of GABAergic reticular units (Yen & Shaw, 2003) suggesting that the somatosensory reticular thalamus may serve as ‘modality gate’ by inhibiting tactile inputs at the same time as letting noxious information pass. Thus, lack of inhibitory GABAergic modulation of spinothalamic input by thalamic interneurons and reticular cells might hypothetically explain why the gating role of sleep spindles on synaptic transmission is not effective for nociceptive information. The enhancement of a late component recorded on the surface at 500–700 ms during spindles was unexpected because such long-latency responses reflect the activation of high-order processing networks linked to the detection of behaviourally relevant input, in both humans (Polich, 2007) and non-human primates (Ueno et al. 2010). Such late ‘P3’ component has been described following nociceptive stimulation in waking subjects, and was considered as an equivalent of the cognitive ‘P300’, or ‘P3b’ wave (Lorenz & Garcia-Larrea, 2003), associated with cognitive closure, memory encoding and stimulus access to consciousness (Polich, 2007). The main contributors to P3 generation are multimodal associative cortices, including the temporo-parieto-occipital junction, as well as the anterior and posterior cingulate and prefrontal areas (Halgren et al. 1998; Br´azdil et al. 2005), and this may explain why it was easily detected on the scalp but not in intra-insular recordings, which may not participate in these high-order networks. P3-like activity has been shown to persist during sleep in response to behaviourally significant stimuli, even if subjects do not remember the stimulus on awakening (Perrin et al. 1999; Bastuji et al. 2003), and its presence was associated with the occurrence of arousals after nociceptive stimuli (Bastuji
12
L. Claude and others
et al. 2008). Thus, an analysis of the relationship between P3 modulation, arousal reactions and spindle should be appealing, although this would have required more stimuli in each condition. In our subjects, P3 enhancement to noxious stimuli delivered during spindling suggests that spindles not only failed to prevent the activation of sensory and associative processing of noxious input, but also enhanced such processing compared to non-spindling sleep periods. Although this view challenges existing ideas on spindle functionality, a possible role of spindling in promoting, rather than depressing, some cortical functions has also been supported by the association between enhanced sleep spindling and improvement of explicit and procedural memory consolidation (Diekelmann & Born, 2010; Fogel & Smith, 2011; Rasch & Born, 2013). These results, together with our own, also concur with data suggesting that sleep spindles may participate in cortical-generated gamma activity (Puig et al. 2008), as well as with experiments in cats showing increased synaptic plasticity during spindles (Timofeev et al. 2002). Sleep spindles are known to be associated with sleep slow oscillations in a dynamic interaction between the thalamus and the cortex (Crunelli & Hughes, 2010) and appear to be preferentially synchronized to the depolarizing slow oscillation phase commonly labelled ‘up-state’ (M¨olle et al. 2011). Failure to disentangle the relationship between spindling and up- or down-states in slow oscillations is clearly a limitation of the present study because up-states are associated with depolarization and vigorous firing, whereas, in down-states, the membrane potential is hyperpolarized and neuronal firing fades. Unfortunately, the sufficient number of stimuli delivered per subject, and especially during N3, needed to investigate the effect of the interaction between spindle and such ultra-slow sleep oscillation on nociceptive responses is lacking in the present data. Two studies, one in humans (Massimini et al. 2003) the other in cats (Rosanova & Timofeev, 2005), showed an influence of slow oscillations on non-nociceptive cortical responses, although the exact role of the combined slow wave/spindle activities on nociceptive input could not be assessed in the present study and remains to be clarified.
Conclusions
Sleep spindles detected in the thalamus or cortex failed to depress arousal reactions, cardiovascular activation or cortical responses to nociceptive stimuli, and could even enhance late associative responses. This may reflect the unique homeostatic relevance of nociception for survival, requiring ‘open access’ to higher centres even during sleep. It also shows that, under particular circumstances, sleep
J Physiol 000.00
spindles do not act as sensory suppressors but may respect or even enhance sensory transmission. References Amzica F & Steriade M (2002). The functional significance of K-complexes. Sleep Med Rev 6, 139–149. Astori S, Wimmer RD & L¨uthi A (2013). Manipulating sleep spindles – expanding views on sleep, memory, and disease. Trends Neurosci 36, 738–748. Bastuji H, Frot M, Mazza S, Perchet C, Magnin M & Garcia-Larrea L (2015). Thalamic responses to nociceptive-specific input in humans: functional dichotomies and thalamo-cortical connectivity. Cereb Cortex doi: 10.1093/cercor/bhv106. Bastuji H, Garc´ıa-Larrea L, Franc C & Maugui`ere F (1995). Brain processing of stimulus deviance during slow-wave and paradoxical sleep: a study of human auditory evoked responses using the oddball paradigm. J Clin Neurophysiol 12, 155–167. Bastuji H, Mazza S, Perchet C, Frot M, Maugui`ere F, Magnin M & Garcia-Larrea L (2012). Filtering the reality: functional dissociation of lateral and medial pain systems during sleep in humans. Hum Brain Mapp 33, 2638–2649. Bastuji H, Perchet C, Legrain V, Montes C & Garcia-Larrea L (2008). Laser evoked responses to painful stimulation persist during sleep and predict subsequent arousals. Pain 137, 589–599. Bastuji H, Perrin F & Garcia-Larrea L (2003). Event-related potentials during forced awakening: a tool for the study of acute sleep inertia. J Sleep Res 12, 189–206. Br´azdil M, Dobs´ık M, Mikl M, Hlust´ık P, Daniel P, Pazourkov´a M, Krupa P & Rektor I (2005). Combined event-related fMRI and intracerebral ERP study of an auditory oddball task. Neuroimage 26, 285–293. Chouchou F, Pichot V, Perchet C, Legrain V, Garcia-Larrea L, Roche F & Bastuji H (2011). Autonomic pain responses during sleep: a study of heart rate variability. Eur J Pain 15, 554–560. Church MW, Johnson LC & Seales DM (1978). Evoked K-complexes and cardiovascular responses to spindle-synchronous and spindle-asynchronous stimulus clicks during NREM sleep. Electroencephalogr Clin Neurophysiol 45, 443–453. Colrain IM (2005). The K-complex: a 7-decade history. Sleep 28, 255–273. Cote KA, Epps TM & Campbell KB (2000). The role of the spindle in human information processing of high-intensity stimuli during sleep. J Sleep Res 9, 19–26. Crowley K, Trinder J & Colrain IM (2004). Evoked K-complex generation: the impact of sleep spindles and age. Clin Neurophysiol 115, 471–476. Cruccu G, Aminoff MJ, Curio G, Guerit JM, Kakigi R, Mauguiere F, Rossini PM, Treede R-D & Garcia-Larrea L (2008). Recommendations for the clinical use of somatosensory-evoked potentials. Clin Neurophysiol 119, 1705–1719. Crunelli V & Hughes SW (2010). The slow (